Chapter 66
Determinism vs Non-Determinism
Non-Determinism
It means: if you send the same request to this black box, you can't be sure you'll get the same answer.
That's why we need to write tons of rules and restrictions — to steer AI into imagining as little as possible.
💡 Setting temperature to 0 does not fix this. It reduces randomness, it doesn't kill it. Floating-point math on GPUs, batching, model updates — there's still enough noise in the system to give you a different answer on a different day. Don't walk into an interview claiming "temperature 0 = deterministic." That's a trap, and a good interviewer will pull the thread.
Part of the full course
Unlock this chapter
This one is part of the paid course. Everything before the interview, the free sampler articles, the Core Java drills, and mock interviews stay open either way.